Daytona, which rents isolated computers to AI agents, says it went from a $1 million to a $100 million run rate in four quarters, according to a 3 November post by its CEO, Ivan Burazin. Modal said on 14 October that its own Sandboxes product is “a $300M run rate product just on its own”.
Together the two sandbox businesses run at about $400 million a year, by our arithmetic. Investor Tanay Jaipuria put the two figures side by side in a post on X on 8 November, writing that “RL workloads seem to be driving the majority with background agents the rest”.
Business Pill · AI Agents Explained
An evergreen Business Engineer explainer on AI agents. It teaches the general idea only and does not describe Modal’s or Daytona’s products.
The key insight: Daytona’s CEO says the company went from a $1 million to a $100 million run rate in four quarters, and Modal says its Sandboxes product alone runs at $300 million, which puts the two agent-sandbox businesses at about $400 million a year, by our arithmetic.
Update (10 Nov): AWS adds another sandbox.AWSannounced AWS Nitro Sandboxes on 9 November: “lightweight VMs managed by the Nitro Hypervisor” that give an Amazon EKS pod, an Amazon ECS task or a lightweight VM, “including AI agents running inside them,” its own hardware-isolated virtual machine. AWS says customers can “safely execute LLM-generated code in an isolated environment of its own” while sharing a single EC2 instance across multiple agentic workloads, and that enabling it in EKS or ECS is “a one-line change.” It is available “on most Nitro-based EC2 instances, in all commercial regions worldwide, at no additional charge.”
Our count: three AWS sandboxes in 16 months. Going by AWS’s own announcements, Nitro Sandboxes is the third AWSproduct built to run agent- or AI-generated code in isolation: (1) Amazon Bedrock AgentCore Code Interpreter, which AWSintroduced in preview on 16 July 2025 as “a secure, sandbox environment so agents can execute code across multiple languages” and which became generally available with AgentCore on 13 October 2025; (2) AWS Lambda MicroVMs, launched on 22 June 2026 to provide “VM-level isolation” for “executing user or AI-generated code,” billed for baseline compute while a MicroVM runs and available in 14 AWS Regions as of 22 October; and (3) AWS Nitro Sandboxes, 9 November 2026. The count and the 16-month span are our arithmetic.
A competitor CEO’s view. Daytona’s CEO Ivan Burazin, whose company sells sandboxes to AI agents and so competes with AWS here, posted on X on 10 November: “AWS launched their 3rd sandbox product today. At this rate theyll have one for every agent…” That is his reading, not AWS’s.
Burazin wrote: “Four quarters. That is how long it took us to go from a $1M run rate to $100M. This year alone, we grew 37.5x.” The rise from $1 million to $100 million is 100-fold, by our arithmetic.
The post says the $100 million figure first showed up as a projection: “Ten days into October, it was already showing $100M. I knew it was a projection.” It adds that “there was still a real chance we wouldn’t finish the month there”.
On what drove it, Burazin described “RL and eval traffic go from almost nothing to a huge share of usage”. He wrote that agent builders “wanted a real computer for every agent: code execution, computer use, reinforcement learning, the messy stuff in between”.
Annualised run rates as stated by each company’s CEO: Daytona’s blog (3 Nov 2026) and Modal’s blog (14 Oct 2026). The $400M combined figure and Daytona’s 100-fold rise are our arithmetic.
Tanay Jaipuria, a partner at Wing VC according to their X bio, posted on 8 November that Modal’s sandbox product is at a $300M run rate and that Daytona went from $1M to $100M in run rate in a year. The post’s view that RL workloads drive the majority, with background agents the rest, is the investor’s own reading.
What Modal Says
In its 14 October post, Modal said it had “crossed $1B in annualized run rate” and raised a “$750M Series D at a $15.75B post-money valuation”, led by Accel. It said it launched Sandboxes in 2023 and that “use cases range from background agents, to reinforcement learning, vibecoding platforms, and much more”.
Modal also said it has “spun up close to 3 billion sandboxes”. Because Modal says it has crossed $1 billion overall, the $300 million Sandboxes business is at most about 30% of its run rate, by our arithmetic.
The Valuations Behind the Run Rates
Daytona raised a $60 million Series B at a $1 billion valuation led by FirstMark, Fortune reported on 20 October. That valuation is 10 times the $100 million run rate Daytona now cites, by our arithmetic. Burazin told Fortune that Daytona’s run rate grew 25-fold between January and September.
Modal’s $15.75 billion valuation is at most 15.75 times its run rate of more than $1 billion, by our arithmetic.
As we read the two posts, the companies are describing the same demand from different sizes: Modal says Sandboxes is one product inside a company past $1 billion, while Daytona describes sandboxes as its whole business.
As we read Daytona’s post, the workload mix is shifting: its CEO writes that RL and eval traffic went from almost nothing to a huge share of usage, which matches the split investor Tanay Jaipuria described on X.
Ivan Burazin, CEO of Daytona, 3 November 2026
“Four quarters. That is how long it took us to go from a $1M run rate to $100M.”
Three Implications
FOR AGENT BUILDERS Daytona’s CEO told Fortune it can configure a new sandbox for an agent in 60 milliseconds; Modal says its sandboxes scale to a million concurrent.
FOR INVESTORS Daytona’s $1 billion valuation is 10 times its stated $100 million run rate, by our arithmetic, per Fortune’s 20 October report.
FOR THE MODEL LABS Both companies name reinforcement learning among sandbox uses; neither breaks out its share of revenue.
The framework says: “The agentic surface — the harness — is becoming where value pools in the entire AI stack.”
As we read it, sandboxes sit underneath that surface: every agent that runs commands or code needs an isolated computer, and Modal and Daytona now report run rates for selling exactly that.
What Is Not Established
Neither company breaks out how much of its sandbox revenue comes from reinforcement learning versus agents in production; Jaipuria’s split is the investor’s own reading. Daytona’s post describes its October figure as a projection first seen ten days into the month. Run rates annualise recent revenue and are not audited results.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Daytona says it reached a $100 million run rate in four quarters, and Modal says its Sandboxes product runs at $300 million on its own, putting the two companies’ agent-sandbox businesses at about $400 million a year, by our arithmetic.
A note on sourcing. Run-rate figures are as stated by Daytona’s CEO (3 November 2026) and Modal (14 October 2026) on their own blogs and are not audited. Valuation and January-to-September growth figures are from Fortune’s 20 October 2026 report. The split between reinforcement learning and agent workloads is the stated view of investor Tanay Jaipuria. Combined and ratio figures are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
